9800GTX+ launched !

Gotta love competition.. Even when the 9800GTX+ was announced, it was going to launch at $229, but no doubt thanks to the 4850 and 4870, it will now launch at $29 less than that and probably drop to $180-$190 within a few weeks :)
 
This has been floating around today. Needs to drop about $50, then is extremely good competition for the 4850. I really can't see these cards going much lower than $150 for the next several months. I know that there are some deals for the 4850 that are a little lower than that, like $140.

Let's hope nVidia can get this going. That's good for the consumer!
 
Why is this thing so huge? They should have been selling these monsters back in 90s..
 
They were included in the benchmarks for most of the 4850/4870 reviews, so they may not get as many of their own dedicated reviews as the benchmarks will all be the same (9800GTX+ vs 9800GTX vs 8800GT vs 4850 vs GTX260).
 
hmz said:
Why is this thing so huge? They should have been selling these monsters back in 90s..
Click to expand...

Many prefer big dual-slot coolers that exhaust the hot air from the case... but it does make the 4850 appear more "sleek" and possibly more optimized (it does put out less heat and only requires a single PCI-E plug).
 
It would be pretty silly to buy this over a ~$150-160 4850. If the price drops by $60 it would be a decent competitor, but at the moment there's really no point.
 
Is there any Benchmarks posted anywherE? If so can someone post the links please, thanks!
 
So what the hell is the difference between it and the normal 9800GTX?!?! They're the same damn card, but the + has a slight OC on the GPU. Even the memory and SP are EXACTLY the same.

If I'm missing something, please tell me. It's just an OC'ed version of the normal GTX. Woopdeedoo. :rolleyes:
 
Red Falcon said:
So what the hell is the difference between it and the normal 9800GTX?!?! They're the same damn card, but the + has a slight OC on the GPU. Even the memory and SP are EXACTLY the same.

If I'm missing something, please tell me. It's just an OC'ed version of the normal GTX. Woopdeedoo. :rolleyes:
Click to expand...

The GTX+ uses a 55nm die as opposed to the original G92's 65nm die. This allows it to run cooler and clock higher, which is why nVidia was able to make this card with such high clocks in the first place. The idea is that the higher clock speed will help it compete with the 4850. That's the difference between this card and the regular GTX.
 
Red Falcon said:
So what the hell is the difference between it and the normal 9800GTX?!?! They're the same damn card, but the + has a slight OC on the GPU. Even the memory and SP are EXACTLY the same.

If I'm missing something, please tell me. It's just an OC'ed version of the normal GTX. Woopdeedoo. :rolleyes:
Click to expand...
Its 55nm
 
Red Falcon said:
So what the hell is the difference between it and the normal 9800GTX?!?! They're the same damn card, but the + has a slight OC on the GPU. Even the memory and SP are EXACTLY the same.

If I'm missing something, please tell me. It's just an OC'ed version of the normal GTX. Woopdeedoo. :rolleyes:
Click to expand...


55nm versus 65nm FAB process (even the 260/280 is 65nm). This is nVidia's first GPU to use it. Less heat, lower power requirement. In theory, even higher OC.
 
Damn, it took a long time for my reply to post. You guys weren't even there when I responded.
 
cannondale06 said:
$199 is not all that great. when it hits $150-$160 then it will be more competitive with the 4850.
Click to expand...
By the time it moves into the $150 territory, the ATI cards will be selling in the $120-130 range.
ATI has got Nvidia by the balls this time around.
 
meh, with the 260 and 280 at the current price points I don't see why they did this at all. like said above they need a 25-35% price drop to make them competitive.

I wonder that they have dropped the 280 and the 260 downs so much and then launched these at this price.
 
There's still the "vanilla" 9800GTX, though. It's nearly as fast as the GTX+, can probably overclock to GTX+ levels or more, and is quite a bit cheaper.
 
Zero82z said:
The GTX+ uses a 55nm die as opposed to the original G92's 65nm die. This allows it to run cooler and clock higher, which is why nVidia was able to make this card with such high clocks in the first place. The idea is that the higher clock speed will help it compete with the 4850. That's the difference between this card and the regular GTX.
Click to expand...
cannondale06 said:
Its 55nm
Click to expand...
Brian48 said:
55nm versus 65nm FAB process (even the 260/280 is 65nm). This is nVidia's first GPU to use it. Less heat, lower power requirement. In theory, even higher OC.
Click to expand...
FIFTY FIVE NANONMETERS

:p
 
Does power consumption go up or down with a die shrink? I thought the smaller lines require higher current?
 
Zero82z said:
It would be pretty silly to buy this over a ~$150-160 4850. If the price drops by $60 it would be a decent competitor, but at the moment there's really no point.
Click to expand...
i agree
 
hmz said:
Why is this thing so huge? They should have been selling these monsters back in 90s..
Click to expand...

What good is a single slot card that runs as hot as an oven? Some people don't mind, personally I don't want that shit in my case.
 
So wait, what exactly is the difference between this and my 8800gts 512mb? (aside from the 55nm die, that is)

Probably not worth upgrading, right?
 
m3ta1head said:
So wait, what exactly is the difference between this and my 8800gts 512mb? (aside from the 55nm die, that is)

Probably not worth upgrading, right?
Click to expand...
no difference other than the shrink as for as the core goes. the 9800gtx and gtx+ boards both have better circuitry. the memory they are using can oc better too.
 
Zero82z said:
You can't really get two 8800GTs for $200, at least not yet. The better buy would be a Radeon HD4850 at the moment.
Click to expand...

especially with the OMFGGREATDEAL!!!111UnOuNO!!!11iCHIi1!1 in the [H]ot Deals forum, $154.99 for a 4850 AND 2GBs DDR2 6400...? wow, but damn, I hate rebates...

:D:D:D
 
I wonder if the speed difference between the 9800GTX and the 9800GTX+, would be a good indicator if the speed difference between the 280, and the new 55nm 280 when released?
 
TalonMan said:
I wonder if the speed difference between the 9800GTX and the 9800GTX+, would be a good indicator if the speed difference between the 280, and the new 55nm 280 when released?
Click to expand...

They're very different GPUs, so I don't think the two die shrinks could be comparable in terms of their effects. It'll probably be more similar to the 90nm to 65nm shrink between G80 and G92 in the sense that the GPU was changed up a bit at the same time.
 
Zero82z said:
They're very different GPUs, so I don't think the two die shrinks could be comparable in terms of their effects. It'll probably be more similar to the 90nm to 65nm shrink between G80 and G92 in the sense that the GPU was changed up a bit at the same time.
Click to expand...

90 to 65nm wasn't really much of a change in terms of IPC or features, except for the addition of a HD video decoder chip onto the GPU itself. that was a die shrink. 65nm to 55nm is an optical shrink, its less substantial, but it yields some improvements in power usage, but that's the only real difference, and possibly a better overclock (legit reviews got an early sample to 855mhz on the stock cooler at stock fan speeds if I remeber correct).

Normally when a company does a shrinkdown of a product, they typically add features, features which make the product itself better and faster. Last few times with nvidia and ATi it doesn't seem that way. ATi went down to 80nm on the same basic x1900 series architecture, then they produced the 2900xt which was a new architecture which was 80nm, then they released the 55nm 3870 which had some fixes, but nothing you would call game changing, they just made the chip work. And you got the 4870, using the same 55nm process. Same thing goes with nvidia and the G80 to G92 transition.
 
Varmint said:
Does power consumption go up or down with a die shrink? I thought the smaller lines require higher current?
Click to expand...

Power consumption goes DOWN with a die shrink.

You're thinking of conductors. If you make a conductor thinner, it increases the resistance (losses) and decreases the speed with which it can charge transistors. And yes, unfortunately, wires must shrink as feature sizes shrink.

How they get around this is clever: they build the wires upwards, so while the wires are extremely thin in length and width, they are incredibly tall. This prevents losses from thinner wires, although they make semiconductor manufacturing difficult.

http://www.research.ibm.com/journal/rd/516/curra3.gif

Transistors themselves require less power when shrunken, because you reduce the capacitance of each gate when you reduce the area. Less capacitance to charge means faster switch times, and less power consumed.

NOTE: the 9800GTX+ does not actually use less power, according to all the reviews I saw. This means that Nvidia is raising the voltage in order to clock the part higher, which really kills any power benefits.

See here:

http://www.pcper.com/article.php?aid=580&type=expert&pid=7
 
Simpson5774 said:
Normally when a company does a shrinkdown of a product, they typically add features, features which make the product itself better and faster. Last few times with nvidia and ATi it doesn't seem that way. ATi went down to 80nm on the same basic x1900 series architecture, then they produced the 2900xt which was a new architecture which was 80nm, then they released the 55nm 3870 which had some fixes, but nothing you would call game changing, they just made the chip work. And you got the 4870, using the same 55nm process. Same thing goes with nvidia and the G80 to G92 transition.
Click to expand...

There weren't any major architectural changes between G80 and G92, but the die was restructured a bit. The number of ROPs were decreased, texture units were increased, and the memory controller was changed to a 256-bit design. I do agree with you mostly, but my point was that I don't think GT200b will be a straight shrink to 55nm, and that it'll probably have a few tweaks similar to what G92 had as opposed to the 55nm 9800GTX+, which is the same chip only on a smaller process.
 
Zero82z said:
There weren't any major architectural changes between G80 and G92, but the die was restructured a bit. The number of ROPs were decreased, texture units were increased, and the memory controller was changed to a 256-bit design. I do agree with you mostly, but my point was that I don't think GT200b will be a straight shrink to 55nm, and that it'll probably have a few tweaks similar to what G92 had as opposed to the 55nm 9800GTX+, which is the same chip only on a smaller process.
Click to expand...
16 ROPs for 256-bit is the same as 24 for 384 and 20 for 320-bit so nothing really changed architecturally in that regard.
 
You must log in or register to reply here.
Back
Top